PodcastMix

PodcastMix是由音乐技术集团和Dolby Laboratories创建的一个大型多样性数据集，旨在解决播客中背景音乐与前景语音分离的问题。该数据集包含约44,455条高质量的语音文件和19,370首音乐文件，均采用Creative Commons许可。数据集的创建过程涉及程序化生成播客内容，确保了数据的真实性和多样性。PodcastMix的应用领域包括提升播客播放时的音乐音量个性化设置，以及改进指纹识别算法和语音分析工具，从而增强播客内容的推荐系统和可视化分析。

PodcastMix is a large-scale, diverse dataset developed by the Music Technology Group and Dolby Laboratories, aiming to solve the problem of separating background music from foreground speech in podcasts. This dataset includes approximately 44,455 high-quality speech files and 19,370 music files, all licensed under Creative Commons. The dataset creation process involved programmatically generating podcast content, which guarantees the authenticity and diversity of the collected data. Potential application scenarios of PodcastMix include personalized music volume adjustment during podcast playback, as well as optimizing fingerprint recognition algorithms and speech analysis tools, thereby enhancing podcast content recommendation systems and visual analytics capabilities.